In [1]:
import sys
import os
sys.path.insert(0, os.path.abspath('../'))
sys.path.insert(0, os.path.abspath('../../'))
sys.path.insert(0, os.path.abspath('/home/hm-tlacherm/qlm_notebooks/notebooks_1.2.1/notebooks/master_thesis_qaoa/'))
sys.path.insert(0, os.path.abspath('/home/hm-tlacherm/qlm_notebooks/notebooks_1.2.1/notebooks/master_thesis_qaoa/ibm/'))
In [2]:
import numpy as np
import qiskit
provider = qiskit.IBMQ.load_account()
from qiskit import Aer
from qiskit.utils import QuantumInstance
from qiskit_optimization.algorithms import MinimumEigenOptimizer
from qiskit.algorithms import QAOA
from shared.QiskitMaxcut import *
from ibm.ibm_parameters import *
from matplotlib import pyplot as plt
%matplotlib inline
In [3]:
graph = generate_butterfly_graph(with_weights=True)
max_cut = Maxcut(graph)
max_cut_qubo = max_cut.to_qubo()
print(graph.name)
max_cut.draw()
In [4]:
step_size = 0.06
a_gamma = np.arange(-np.pi, np.pi, step_size)
b_beta = np.arange(-np.pi, np.pi, step_size)
In [5]:
len(a_gamma)
Out[5]:
In [6]:
a_gamma, b_beta = np.meshgrid(a_gamma, b_beta)
In [7]:
quantum_instance = QuantumInstance(
backend=Aer.get_backend(DEFAULT_QASM_SIMULATOR),
shots=SHOTS)
qaoa = QAOA(
optimizer=COBYLA(maxiter=0),
quantum_instance=quantum_instance,
reps=1
)
op, offset = max_cut_qubo.to_ising()
In [8]:
def maxcut_obj(x, G):
"""
Given a bitstring as a solution, this function returns
the number of edges shared between the two partitions
of the graph.
Args:
x: str
solution bitstring
G: networkx graph
Returns:
obj: float
Objective
"""
obj = 0
for i,j,w in graph.edges.data('weight'):
if x[i] != x[j]:
obj -= w
return obj
def compute_expectation(counts, G):
"""
Computes expectation value based on measurement results
Args:
counts: dict
key as bitstring, val as count
G: networkx graph
Returns:
avg: float
expectation value
"""
avg = 0
sum_count = 0
for bitstring, count in counts.items():
obj = maxcut_obj(bitstring, G)
avg += obj * count
sum_count += count
return avg/sum_count
In [9]:
def create_cirucit(beta,gamma):
circuits = qaoa.construct_circuit([beta,gamma], operator=op)
circuit = circuits[0]
circuit.measure_all()
return circuit
In [10]:
landscape = np.zeros(a_gamma.shape)
for i in range(0, len(landscape)):
circuits = []
for j in range(0, len(landscape)):
# create circuits for entire row
circuit = create_cirucit(b_beta[i][j], a_gamma[i][j])
circuits.append(circuit)
# create one job with circuits
job_name = f"{graph.name}_row_{i}"
job = qiskit.execute(circuits, backend=provider.get_backend('ibmq_toronto'), shots=1024)
job.update_name(job_name)
print(job_name)
# add results to landscape
j = 0
for count in job.result().get_counts():
mean = compute_expectation(count, graph)
landscape[i,j] = mean
j += 1
In [11]:
print(landscape)
plt.matshow(landscape)
plt.show()
In [12]:
# Mean of landscape
np.mean(landscape)
Out[12]:
In [13]:
# Minimium
np.min(landscape)
Out[13]:
In [14]:
# Display Coordinates of Minimum
np.unravel_index(np.argmin(landscape), landscape.shape)
Out[14]:
In [15]:
# Gamma and beta value of Minimium
gamma, beta = np.unravel_index(np.argmin(landscape), landscape.shape)
opt_gamma = gamma * step_size
opt_beta = beta * step_size
print(f"Opt.Gamma: {opt_gamma}, Opt.Beta: {opt_beta}")
In [16]:
# Save result matrix
with open('landscape_toronto_butterfly_weights_results.npy', 'wb') as f:
np.save(f, landscape)
In [17]:
import plotly.graph_objects as go
In [18]:
# Plot landscape in 3D
a_gamma = np.arange(0, np.pi, step_size)
b_beta = np.arange(0, np.pi, step_size)
fig = go.Figure(data=go.Surface(z=landscape, x=a_gamma, y=b_beta))
fig.update_traces(contours_z=dict(show=True, usecolormap=True, highlightcolor='limegreen', project_z=True))
fig.update_layout(title="QAOA MaxCut", scene=dict(
xaxis_title="gamma",
yaxis_title="beta",
zaxis_title="mean"
))
In [19]:
# Plot Heatmap
fig = go.Figure(data=go.Heatmap(z=landscape, x=b_beta, y=a_gamma, type = 'heatmap', colorscale = 'viridis'))
# Update Layout
fig.update_layout(title="QAOA MaxCut", width=700, height=700, xaxis_title="beta", yaxis_title="gamma")
# Display Global Minimium
fig.add_trace(
go.Scatter(mode="markers", x=[opt_beta], y=[opt_gamma], marker_symbol=[204], text = [landscape[gamma,beta]],
marker_color="red", hovertemplate="x: %{x}<br>y: %{y}<br> z: %{text:.2f}<extra></extra>",
marker_line_width=1, marker_size=16))
In [20]:
# Display Optimizer Results
# Display path
#fig.add_trace(
# go.Scatter(mode="lines", x=gammas, y=betas, marker_symbol=[200],
# marker_color="white", marker_line_width=1, marker_size=8)
#)
# Display start point
#fig.add_trace(
# go.Scatter(mode="markers", x=[gammas[0]], y=[betas[0]], marker_symbol=[204],
# marker_color="gray",
# marker_line_width=1, marker_size=16))
# Display end point
#fig.add_trace(
# go.Scatter(mode="markers", x=[gammas[-1]], y=[betas[-1]], marker_symbol=[204],
# marker_color="green",
# marker_line_width=1, marker_size=16))
In [21]:
# Plot Optimizer History
#fig = go.Figure(data=go.Scatter(x=counts, y=values))
#fig.update_layout(xaxis_title="Evaluation Counts", yaxis_title="Evaluated Mean", title="Optimizer")
#fig.show()
In [ ]: